Fuzzy diagnosis method based on multi-frequency responses for tolerance analog circuit

For tolerance analog circuit, a novel fuzzy diagnosis method fusing the information of signal features of multi-frequency responses is proposed. The AC signals with different frequencies are respectively input to the circuit for Monte-Carlo analysis. The response in time domain are modeled by the wave description model as so-called feature interval vector. The waveform similarity is defined as the similarity of the test sample to the feature interval vector. Then, multiple inputs with different frequencies bring about several responses and waveform similarities, by which the overall similarity is calculated through information fusion and finally, a consistent diagnosis conclusion is drawn. The experimental results show that the proposed method, dealing with the tolerance in the analog circuit efficiently, leads to more accurate diagnosis results and reduces the test cost. It is effective even in the slight soft fault diagnosis.